Development of a 1-year risk-prediction nomogram for good functional response with anti-VEGF agents in naive diabetic macular oedema.

ABSTRACT

PURPOSE: To develop a risk-prediction nomogram based on baseline variables for good functional response during the 1st year of treatment with anti-VEGF agents in naive diabetic macular oedema (DME). METHODS: This retrospective study included patients presenting naive-DME treated with anti-VEGF therapy at Dijon University Hospital (France) between 1 February 2012 and 31 March 2015 (derivation cohort). We studied baseline variables that had significant associations with a good functional response to anti-VEGF agents during the 1st year of treatment. We used a program to generate a nomogram based on a binary logistic regression predictive model. Then, this nomogram was tested on data from a separate cohort of naive-DME patients from a multicenter study involving 20 French ophthalmologic centres between January 2014 and June 2015 (validation cohort). RESULTS: Age, baseline BCVA and ellipsoid zone integrity on spectral-domain optical coherence tomography (SD-OCT) are functional prognostic factors and were used to build a nomogram. The nomogram showed excellent discrimination for good functional responders (area under the curve (AUC) = 0.906, 95% confidence interval (95% CI) = [0.849-0.964], p = 0.004). The discriminative power of this nomogram was tested on the validation cohort data, demonstrating good discrimination of good functional responders (AUC = 0.942, 95% CI = [0.898-0.986], p < 0.001). CONCLUSION: This nomogram provides a useful estimation of a good functional response in naive-DME patients treated with anti-VEGF agents.